Effectiveness of the canvass in 2023

At the beginning of the canvass, all registers are matched against Department for Work and Pensions (DWP) data. The results are used by EROs to allocate properties to ‘routes’ which determine how many times they will contact a household to attempt to get a response. In Route one households get fewer contacts, in Route two they get more (see Background section below for more detail).

EROs can also carry out further matching using local data, such as council tax records. The majority of EROs carried out local data matching and, as in previous years, council tax records were the most popular datasets used.

The local data matching step had a significant impact on the number of households allocated to each route, as shown in Table 7. When  asked whether their approach to local data matching was the same as or different to last year’s, a majority (85%) of those EROs who responded said that their approach was the same. As in 2022, for those that decided to take a different approach, the responses suggest that this predominantly involved using a wider range of data sources. In a small handful of cases, EROs used fewer datasets in 2023 relative to 2022. These EROs cited concerns around the quality of local data, difficulty using the data matching function within their electoral management software (EMS) system, and resource shortages as reasons for using fewer datasets this year. Other changes that EROs made to the data matching process in 2023 included allowing more time to complete the data matching step and doing the data matching step at a later stage in the process.

Across the board, those local authorities who did not conduct local data matching allocated fewer properties to Route 1 (68% vs 76%), and more properties to Route 2 (31% vs 22%).

As shown in Table 8 (below), the percentage of properties allocated to Route 2 was (24.9%). This continues to be in line with the expectations set out in the statement of policy for the reformed canvass published by the UK, Scottish and Welsh governments in 2019, which predicted that approximately one quarter of properties would need to be allocated to Route 2 nationally2 .

The proportion of route allocations continues to be consistent across Great Britain, with the exception of London. London continues to have a lower percentage of Route 1 properties (66.6%) and a higher percentage of properties allocated to Route 2 (32.1%) in comparison to other English regions. This reflects a higher rate of population movement in London with fewer properties successfully matched against existing DWP and local data due to the rate of change in household composition.

The changes to the annual canvass were designed to allow EROs to direct their resources at households whose composition was most likely to have changed. The effectiveness of the canvass, based on this targeting of activity, will therefore be strongly determined by two things: the accuracy of the data matching and the level of response achieved. 

As expected, the response rate is significantly higher among Route 2 households, where changes are expected, than Route 1 households, where they are not (65.5% vs 24.5%) (see Table 9).

In 2019, before the canvass process was reformed, 23.8 million household responses were received during the canvass. In 2023, there were 10.3 million responses received, which is on par with 2022 and only slightly higher than the level of response in 2020 and 2021 (~9 million). The drop in responses compared to 2019 can partly be explained by the lower response rate among Route 1 households, which comprise the vast majority of households and who are now not required to respond unless there has been a change in the household composition.

However, as in 2021 and 2022, there are still around a third of households in Route 2 who did not respond, despite changes being expected in these properties. Our accuracy and completeness research did not find a decline in the accuracy and completeness of the registers, however there was also no improvement. Higher rates of response among Route 2 households would be needed to improve accuracy and completeness.

Several EROs suggested that funding pressures made it difficult to achieve a higher Route 2 response rate, in part because these pressures made it harder to recruit canvassers. Others were unable to run public awareness campaigns or were impacted by the increase in the cost of postage. A growing number of EROs are using e-communications and telephone canvassing as this represents better value for money.  This presents some limitations as telephone data is not always complete and respondents can be wary of e-communications from the ERO and mistake them for spam, however there are encouraging signs that e-communications are becoming more useful. Some EROs have reported that the number of email addresses on record are increasing, and people are increasingly less wary of receiving e-communications from their local authority.

It continues to be a challenge to reach communities who have historically been reluctant to engage with canvassers. Densely populated urban areas with high rates of population movement appear to be the most challenging areas. Some EROs are choosing to target their door-knocking efforts only in these locations. However, data from previous years demonstrates that even that is not universally successful.

We will focus our analysis of the Route 2 data over the coming years to further develop our understanding of the challenges faced by EROs and to support the identification of actions to help mitigate against the risks to the accuracy and completeness of their electoral registers. We will also undertake work with smaller groups of EROs to facilitate the sharing of good practice between those facing similar challenges.

As in all years since 2020, Scotland continues to have the lowest response rate amongst route 2 properties (50.4%) compared to 64.0% in Wales and 67.2% in England (Table 9). However, despite the difference in response rate, the latest accuracy and completeness study did not find that the completeness nor accuracy of the registers in Scotland had declined since 2018.

Route 2 response rates in England remain at roughly the same level as they have been for the past three years, though they have declined since the peak in 2021 (Table 10). Route 2 response rates in Scotland and Wales have remained steady over the last four years.

The nature of the responses received can also give an indication of the efficacy of the reformed canvass process. Responding households can record a major change (e.g. reporting that a potential new elector is resident), a minor change (e.g. amending the name of an existing elector) or no change (i.e. confirming the existing details of household members). Understanding the distribution of these changes can tell us about the accuracy of the data matching process.

Of the 10.3 million responses received across all routes, 2.6 million households reported a major change. Table 11 describes how these major changes were distributed across the three routes. 

The pattern is broadly in line with 2020-2022, where approximately two thirds of major changes were reported by Route 2 households – i.e., where one would expect changes to be reported. As in previous years, a third (33.6%) of major changes relate to Route 1 households, where DWP and/or local data had indicated there should be no change needed.

As in the previous two years, of the 22.1 million households allocated to Route 1 (including those who did not respond to the canvass), the percentage reporting a major change (4.0%) is small. However, as Figure 3 shows, over a tenth (16.6%) of those Route 1 households who responded to the canvass reported a major change. Importantly for the quality of the registers, it is also unlikely that all the Route 1 households that needed to report a major change have done so – particularly as they would have received limited contact from the ERO. 

Canvassing households does not directly result in new registrations. When a household reports that a potential new elector is resident, that individual still needs to submit an application to be added to the register. Where a household reports that electors need to be removed from the registers, a second piece of evidence (e.g. locally held data) is needed before the ERO can confirm the deletion.

Entries need to be added to, and deleted from, the registers for several reasons including migration, home movement and deaths. The level of these additions and deletions provides insight into whether registration activity is keeping pace with population change. As population mobility varies across the country, so does the scale of the challenge faced by EROs.

As in 2020-22, Route 2 households account for the largest proportion of additions and deletions (see Table 12). However, a sizeable proportion (around 40%) of additions and deletions came from Route 1 households, whose composition was presumed to be unchanged. Again, this suggests either a degree of inaccuracy in the allocation of households to routes or an impact from population movement between matching and canvassing.

There is at least some population change not being picked up on the registers. Our accuracy and completeness research did not find evidence that the quality of the registers has declined since 2018 before the canvass was reformed, however as it also has not caused significant improvement, this is an area for attention.

Table 13 shows the levels of additions and deletions in recent years across the whole year, not just the canvass period. The percentage of additions and deletions in 2023 are stable compared to 2021 and 2022, which have been slightly lower than previous years. However, it usual to see more change in years with significant UK-wide electoral events, such as a general election or referendum (e.g. 2015 or 2019).

As set out in Table 14, we also expect to see a higher proportion of additions and deletions during the canvass period than outside in years where there is no UK-wide electoral event. This trend continued in 2023.

Table 15 below shows the proportion of additions and deletion recorded during the canvass period across the three nations of Great Britain. The proportion of additions and deletions during the canvass were highest in Wales (65.1% and 70.6% respectively), where there were no elections during the year. Additions were lowest in Scotland (61.2%) while deletions were lowest in England (65.2%).